Advertise here.

social_visualization_benchmark.jpg
Here is the culmination of our previous three posts reviewing the big 3 of social visualization tools: Swivel, Many Eyes, and Tableau Public. This is it - the cage match, the final showdown.

Using the (arguably) peerless example of information visualization, Minard's infographic of Napoleon's march, I endeavored to test the capabilities of each tool, head to head. The goal was to use the multi-layered and data-driven storytelling capabilities of Minard's graphic as a benchmark for visualization.

Setting off, I thought this was going to be very fun process. I was wrong. Very, horribly wrong. I hope the details of this affair will amuse and enlighten.

Step 1: Start off easy.
My first thought was to start with the seemingly simplest tool, Swivel. SImple, yes, but painful when using any real world data. I knew Swivel does not do cartography, yet I had visions of displaying this information in other novel ways. Unfortunately, the tool is so inflexible in its formatting and display options, that we eventually gave up on it.

The cool part is that you can create reports. The bad part is that you cannot embed a report.

Here is one of our charts. It shows the survivors over the longitude values. Unfortunately, we couldn't figure out how to show each troop group separately.

Here is the full report on Swivel.

Step 2: Kick it up a notch.
Many Eyes beats out Swivel with its plethora of visualizations. I had high hopes for mapping out the data. However, locations aren't supported in their world map; it can only be used as a choropleth map.

So again we set off to visualize the data non-spatially. I was hoping to show a network graph of the paths of the three groups of troops, but the options were so limited, that it wasn't useful. The node size and color cannot change based on variables.

We finally decided on a treemap, scatterplot, and bar graph. The scatterplot shows conceptual spatial information (not based in reality of lat/long). It at least gives a sense of north/south/east/west. The treemap (below) shows the vast numbers advancing (white) and the meager numbers retreating (orange). Finally, the bar graph shows the temperatures.

See all of the Many Eyes examples in the Minard Topic Center.

Step 3: Hit it out of the park.
By this point, I am realizing that Tableau Public is really going to put these other tools to shame. I was able to quickly create this pretty thorough replica of Minard's graphic. If I had spent more time, I could have "evolutionized" it to show more of the data and in different ways.

It looks better at full size.

So, you be the judge. Is the Minard graphic even a fair or useful benchmark to use? Is there a better data set or concept to use as our yardstick?

Should these tools even be compared to each other? They have different pros and cons and use cases. Is it worthwhile to compare them as equals?

Reviewed by Kim Rees and Domanique Alicia of Periscopic, a socially-conscious Information Visualization firm specializing in helping nonprofit organizations and like-minded companies convey important messages and elevate public awareness.

11 COMMENTS

Excellent work and a really good test of the flexibility and versatility of tools such as those used. The Minard graphic probably isn't a fair test given the multi-variate nature of the challenge it poses. However, I'm sure Tableau will be very happy given that your attempt using it largely nails it. I think ideally there would be a single tool to serve all needs, but the reality (at least right now) is that a suite of solutions at least give you different options, utilities and provide alternative strengths. The important thing is for users to be able to clearly understand the pros/cons and scope of each in order to efficiently decide which one to use to embark on tasks like this. Your previous reviews of such resources give a great help to achieve this.
Its strange that you have posted this today given the Gene Zelazny essay, re. Minard's graphic, which I talk about here: http://www.visualisingdata.com/index.php/2010/06/napoleons-moscow-march-never-seems-to-end/
Cheers
Andy Kirk

Wed 23 Jun 2010 at 10:48 PM

You missed checking out GeoCommons - which in particular if focused on superb mapping capabilities and upcoming charting. Line thematics soon :)

Andrew
http://geocommons.com

Thu 24 Jun 2010 at 12:26 AM

Thanks for the tip, Andrew. We're thinking about using this benchmark for ongoing reviews of tools, libraries, etc. I'll add GeoCommons to the list!

Thu 24 Jun 2010 at 5:26 AM

Andy, that is strange! I read Andrew Arbela's post yesterday and thought the same thing. :) Thanks for linking to your response. I agree with your points. Great retort!

Thu 24 Jun 2010 at 11:59 AM

My take on this viewing it from iPhone. Pretty stunning differences of quality and insights, whether or not you view it on iPhone. https://twitter.com/Freakalytics

Thu 24 Jun 2010 at 12:29 PM

Thank you for your visualizations.

To compare or not compare, that is the question! To be honest I have no idea, Everything depends on your skills, data, expactations and so on. I use mainly maps like this map of Russia http://manyeyes.alphaworks.ibm.com/manyeyes/visualizations/population-of-russian-regions (as I am blogging about Russia :) so I am happy with ManyEyes. ManyEyes is online software, unfortunately I have tendency to download too many software that I later do not use.

I also downloaded Tableau but to create maps of Russia I would have to instal the map, as I am not advanced and a bit lazy, ManyEyes of Russian regions is really tempting to me :)

Thu 24 Jun 2010 at 7:26 PM

Thanks for your examples, Russian Sphinx. It looks like Many Eyes works well for your data. The frustrating thing for me was that I couldn't show a path or locations spatially on their maps. They really only do choropleth. Which is great and actually lacking in Tableau. So perhaps the softwares complement each other in their mapping capabilities.

Fri 25 Jun 2010 at 5:05 AM

Thank you very much. Now, I´m using tableu public for my personal project.

Mon 28 Jun 2010 at 11:11 AM

Kim,
Great review - your Tableau Minard map is simply gorgeous. Is it fair to use this as your benchmark? Possibly not, particularly as swivel doesn't really make any claims to be a mapping tool. Trying to reimagine such a famous chart as Minard's using regular bar charts is slightly unfair on Swivel.

I wonder if there are any other famous charts that could be used as benchmarks?

The output, of course, is not the only point in using a benchmark - it's also a way of testing the process of creating and embedding a viz, and this clearly establishes Tableau as the winner.

Great work - your reviews of these products have been really interesting to read.

Andy

Thu 08 Jul 2010 at 8:25 PM

Great review! It seems that Tableau is the most flexible in dealing with a complex dataset (which is what I would expect). I'm not sure if this dataset is a fair benchmark, it depends: if your point is to show that Tableau has more capabilities in dealing with complex datasets, it is. If your point would be to show the differences in the way each tool handles a certain kind of dataset it can handle, you should probably pick a different dataset.

Anyway, a very good review and fun to read! :)

Thu 08 Jul 2010 at 8:40 PM

Thank you for this review! Recently I was able to gain first in-depth experiences with Tableau. Read my summary (german) here: http://www.wangenrot.at/2010/293-tableau/

Tue 21 Sep 2010 at 1:54 AM
ADD A COMMENT
Commenting has been temporarily disabled.